The internet that I knew and grew up with is dead and has been that way for a while. To me the downfall started when everything online became a financial pursuit in some shape or form. However, even then, it was possible to find pockets of genuine creativity and discussion. The advent of LLMs was the final blow, and the reason is exactly the fact that one can't assume authenticity in anything these days.
Yes proof-of-identity and authorship would help, however that comes with the extremely perverse side effect of invasive identity verification. Sadly zero knowledge proofs for that are still brittle and not well developed. Hopefully something like that is able to materialize without being the privacy nightmare that is the myriad identity verification mechanisms currently mandated by several governments.
Same here, first time on the Web 1995, a few BBS access before it.
For hobby projects coding with LLMs feels like fast food.
When someone posts a project with AI collaboration, I know from my own work, that they only commissioned the project to agents, just like we commission someone else to deliver a house, a painting, a ready made piece of furniture, whatever.
I can clearly remember the 90ies, when it wasnt usual to "be always online", mainly due to limited connections & providers and huge costs in our regions. I can also remember they years when I didnt had "a somewhat permanent" internet connection at home and when I had to pick up my computer, drive it to a friend and connect there to do some downloads :-D
I miss the old forums esp., nowadays everybody posts on LinkedIn and its the "main online place to be"
I don't think the only way is the ID verification. That's what the spying side pushes as a narrative, but there are ways to validate each other without global registries and any government connections.
Web-Of-Trust was a good idea with poor implementation way ahead of its time. Closed communities work pretty well and seem to be almost unaffected.
So yes, the public internet is dead but I believe that it will be solved with some sort of reputational/vouch systems and invite-only groups where you have to contribute to be accepted and have an incentive to protect that seat instead of posting slop/spam/other agenda to be banned.
I think even pre-LLMs, the problem was more the size of a community rather than whether it was closed. The bigger it grows, the more it orients itself towards the lowest common denominator. Every community I ever cared about was rather small. Places like Hacker News, which I think is among the better large ones, I honestly don't think I'd care much if it disappeared tomorrow.
But now with AI unfortunately I agree. Not sure if there is a way around closing off communities. Even for communities that actively use AI within.
99% of the people are in read-only anyway. The community can be closed for contributions, but open for reading. That might cut bots, ads, propaganda and espionage, but retain the sharing function for those who need without losing much.
The bigger the "community" is, the less it's actually a community. HN is not a community but a social media feed, e.g. like Instagram but with text instead of video. Lobste.rs might be a community, I don't know because I don't have an invite. Most physical hackerspaces (not makerspaces) remain communities even as they get thousands of members.
The moment you assume someone creates it, you have a problem. It has to arise organically out of the individual actions of all of us.
I think of the "fediverse" as an example of this. It's usually defined as the ecosystem around the ActivityPub protocol (i.e. Mastodon, Lemmy, etc) but I think it should be redefined more broadly to mean "the ecosystem of federated/decentralised publishing and interaction", which would include ActivityPub but also blogs, websites, email, RSS, ATproto. If we really have a fedi-verse, then it can't be controlled by one entity.
Sorry, it was a loaded question. I was more asking who's working on the foundation for this next phase of the internet. Who's working on decentralized reputational/vouch systems? Who's working on zero-trust proof-of-authorship or proof-of-being? Who's working on privacy-focused proof-of-identity? etc.
But how is that an "internet"? The point of the free internet was to be free to communicate with whomever.
When you seclude yourself in bubbles, you might feel protected, but really you expose yourself to becoming irrelevant.
The crucial problem is, people confuse message with messenger.
When comments are objectively true and faithful (representing the topic correctly), it doesn't matter who wrote them.
The other way around, knowing somebody personally doesn't guarantee their messages to be true and "not spam". The important thing is to check content thoroughly for actual truth and point out deviations. Just like in an ideal scientific discussion. The remaining social part is a question of civil discourse and enforcing that.
Objectively true statements don't depend on any human understanding them for being true. So it doesn't matter who acts as their messenger.
You conflate two disparate things there: you as the reader understand something as a human. Who wrote it plays no role for the true content. Expressions of human sentiment are another thing entirely.
When LLMs are used to astroturf and manipulate human community building, they rely on such sentiment to stir emotional reactions. Without a rational, objectively true, basis to stand on, that remains sterile if people value truth.
The problem we currently see comes from people foregoing that rooting in objective truth entirely. They act on pure sentiment and are prone to automatic manipulation by it.
You seem to be externalizing the cost of verifying that a thing is true. The cost is nontrivial.
Consider for a moment that all of science is, in some way or another, the search for truth. I have a hard time seeing that any one person—or even any one person with an amazing computer program—can substitute for generations of humans trying to achieve that goal. It is extremely difficult to know what is objectively true.
What a weird take on things: you are currently externalizing the non-trivial cost of verifying that a thing is true.
When you want to understand something as true, you are the one who has to verify it, implicitly. People currently are used to others (supposedly) doing that for them. That doesn't mean, it wasn't externalized.
Now, you're right about individuals being overtaxed without the collaborative effort of society. You have to have chains of trust to have any chance, or personalized AI (which again you would have to trust).
But I was talking about the process of filtering truth collectively. You start from what you believe to know now, team up with others sharing your opinions and go from there. Revising your preconceived convictions is part of proper science.
I'm always so disappointed when I see a "Pricing" link on "Show HN" projects. Open source and giving things away was always the most cool/radical part of the web for me.
> Yes proof-of-identity and authorship would help, however that comes with the extremely perverse side effect of invasive identity verification
It doesn't help because there's a powerful incentive for people to, frankly, lie. HN is chock-full of obviously AI-generated articles where the authors, caught red-handed, insist it's not their hand. They want the best of both worlds: all the recognition, the reach, and the opportunities that come with being seen as an online authority, but none of the actual effort that you can now simulate with an LLM.
Real ID makes the stakes a bit more tangible - "if you die online, you die in real life" - except with eight billion people on the planet, there's an inexhaustible supply of people who will create accounts for you for a sweet $100.
I find that PowerShell's syntax choices and verboseness make it just as cryptic as bash. "Okay, now we Get-Object... Ah crap, that's not it, what was it again? Ah yes, Select-Object... Okay now we pass that into Call-WebRequest... Wait that's not it either...".
The language designers managed to introduce the ambiguity of natural language into a shell scripting language. As a developer, I think that it makes PS wildly unergonomic and I find it easier to remember bash/Linux idiosyncrasies.
Introducing natural language uncertainties into a scripting language reminds me of AppleScript, which used almost entirely English words (mainly in sentences or at least verb phrases), but was nonetheless notably definitely not English. It was quite possible to forget at times which English verbs were language keywords and which weren't.
If I remember correctly, AppleScript was inspired by HyperTalk, which also mainly used English keywords in English-inspired syntax, but most HyperTalk wasn't valid AppleScript, and vice versa.
There is a standard[0] you are ignoring either intentionally or out of ignorance. It is okay that you are used to Bash but please stop spreading false information about PowerShell.
It's not false information. I'm just saying it's as idiosyncratic as bash is (and no, defining its own special standard doesn't make it less so, it just makes its idiosyncracy well defined), and that I have a hard time agreeing with anyone who says that PowerShell's syntax is more intuitive than bash. You still need to learn the special naming scheme, same as needing to learn bash's own quirks. Someone who comes from bash to PS might not know about the existence of that "standard", and might fumble about as I described.
It's amazing how much they've lost the tinkerer audience over the last few years. Unless you really need the GPIO, there are far better and cheaper alternatives for a low power computer.
> It's amazing how much they've lost the tinkerer audience over the last few years.
They haven't lost it. Their goal was to kickstart something and they succeeded beyond their wildest dreams. The space has moved - a usable computer now demands a lot more than the original Pi, and they realized that - this is why their tinkerer platform is the RP2040 family. The market is very healthy and the ESP32 family has a huge space.
What RPi brought was the idea that a platform with good software support would allow for a revolution in the space above the power of an Arduino.
You might've skimmed over the rest of my comment, but my focus was on another type of tinkerer - and maybe that was my bad, I guess "lite homelabber" might have been a more apt description. Raspberry Pis used to be huge among people who wanted a low power device to host useful services. I myself was one of those people once. I never used the GPIO pins, but even today a Pi is my DNS (PiHole) and Home Assistant server. It was a great device to serve as an intro to self-hosting, but with where they've taken pricing and as you said, where the market has gone, they've become all but irrelevant in that space.
But again, you're assuming that that's a bad thing. In fact if they were still doing the same thing that they were 14 years ago it would mean that they had failed their mission of helping bring low powered devices to the underprivileged.
They jump started a market that has created dozens of competitors that are competing to provide the best value for the lowest price to the market with their own niche fits and geo locations and sourcing deals, etc, etc.
It’s a bit like saying the OLPC project failed. Their hardware was never ubiquitous, but they lit a fire under the seats of the industry giants that caused things like netbooks and the modern laptops built almost specifically for schools (cheap, indestructible, easy to maintain), and many schools save resources by getting students electronic versions of their books.
I use one of those myself when I don’t want to carry dead weight.
Of course, I’d love if the OLPC project lived up to their promise, but that was the stretch goal.
Even if you want the GPIO the RPI Zero or Pico are much better options. If you want fast computation on top of that simply buy any other computer and send signals to the Zero or Pico for the GPIO pins.
The RPI Zero 2 has micro-USB ports ... there are plenty of reasons given for this, which arguably made sense 5 years ago. I'm not sure about 2026 though.
People are still using Arduino boards with USB-A mini B ports. At least for micro USB you can easily find cables everywhere. Hardware world is kind of slow in that regard. You can always go for ESP32 if you want though. Cheap and USB-C.
Depends on what you want to do. ESP32-S3 devices are ridiculously cheap and come in a myriad of form factors. If you want a RP2040 form factor, they're like $5 each. They've got Bluetooth and WiFi built in, plus plenty of GPIOs.
If you need compute with a bit more horsepower, the myriad of NUCs or thin PCs exist. If you need both in a single board, then the myriad of other SBCs that aren't Raspberry Pi's...
a RPI is just a SBC with GPIO pinning (which isn't at all unique to RPI.).
The main reason it was/is the popular behemoth is not because of it's tinkerability, but because it got mainstream appeal and RaspianOS is polished enough for the average windows user to go "wow, neat."
THere is no "knock off pi", there are many many popular SBCs with the same performance at cheaper cost (O-Pi, arduino), not to mention the mini PC space (NUCS, ie intel asus.)
If you're a hobbyist, pick up some used thin clients from Ebay. I lost a couple of pis recently, and instead of replacing them, I bought a batch of Wyse 3040 thin clients for about $23 each, power supply (which is 12V) included. 2Gb RAM and a 4-core intel Atom - roughly equal to a Pi 4, but at (what is now) 1/3 the price.
Is anyone surprised at this? Abandoning products in an untimely fashion is their specialty. I've never bought a hardware product from them for that reason, except for a 2nd gen Chromecast which was something whose longevity I didn't care too much about.
Yes proof-of-identity and authorship would help, however that comes with the extremely perverse side effect of invasive identity verification. Sadly zero knowledge proofs for that are still brittle and not well developed. Hopefully something like that is able to materialize without being the privacy nightmare that is the myriad identity verification mechanisms currently mandated by several governments.
reply